Changes between Initial Version and Version 1 of AvailableOptionsforInitPDAF


Ignore:
Timestamp:
Feb 13, 2012, 5:15:30 PM (14 years ago)
Author:
lnerger
Comment:

--

Legend:

Unmodified
Added
Removed
Modified

  • AvailableOptionsforInitPDAF

    v1 v1  
     1= Available options for the different filter algorithms =
     2
     3[[PageOutline(2-3,Contents of this page)]]
     4
     5There are different operations for each of the filter algorithms that need to be specified in the call to `pdaf_init`. To display the available options in a compiled assimilation program, one can use run with the specification `subtype=-1`. In this case `pdaf_init` wil display the available options for the selected filter algorithm and exit with an error status of -1.
     6
     7Below we list the operations are they are displayed for `subtype=-1`.
     8
     9== SEEK ==
     10
     11{{{
     12        Available options:
     13
     14        Sub-types (Parameter subtype)
     15          0: Evolve unit modes with finite difference approx. of TLM
     16          1: like 0 with modes scaled by eigenvalues, unit U
     17          2: Fixed basis vectors; variable U matrix
     18          3: Fixed covariance matrix (V and U kept constant)
     19          5: Offline mode
     20
     21        Integer parameters (Array param_int)
     22          param_int(1): Dimension of state vector (>0), required
     23          param_int(2): Ensemble size (>0), required
     24          param_int(3): Interval for re-diagonalization of P (>0); optional: default 1
     25          param_int(4): 1 for incremental updating, 0 else; optional: default 0
     26
     27        Floating point parameters (Array param_real)
     28          param_real(1): Forgetting factor (usually >0 and <=1), required
     29          param_real(2): epsilon for finite-difference approx. of TLM, required
     30
     31        Further parameters
     32          n_modeltasks: Number of parallel model integration tasks
     33              >=1 for subtypes 0 and 1; not larger than total number of processors
     34              =1 required for subtypes 2 and 3
     35          screen: Control verbosity of PDAF
     36              0: no outputs
     37              1: basic output (default)
     38              2: 1 plus timing output
     39              3: 2 plus debug output
     40
     41        +++++++++ End of option overview for the SEEK filter ++++++++++
     42}}}
     43
     44
     45== SEIK ==
     46
     47{{{
     48        Available options:
     49
     50        Sub-types (Parameter subtype)
     51          0: full ensemble integration; left-sided application of T
     52          1: full ensemble integration; right-sided application of T
     53          2: Fixed error space basis
     54          3: Fixed state covariance matrix
     55          4: Implementation with explicit ensemble transformation
     56          5: Offline mode
     57
     58        Integer parameters (Array param_int)
     59          param_int(1): Dimension of state vector (>0), required
     60          param_int(2): Ensemble size (>0), required
     61          param_int(3): not used
     62          param_int(4): 1 for incremental updating, 0 else; optional, default: 0
     63          param_int(5): Type of forgetting factor; optional, default: 0
     64              0: fixed forgetting factor
     65              1: adaptive forgetting factor (experimental)
     66          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     67              0: deterministic omega
     68              1: random orthonormal omega orthogonal to (1,...,1)^T
     69              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     70                 (experimental; for random transformations, 1 is recommended)
     71          param_int(7): Type of transformation matrix square root; optional, default: 0
     72              (Only relevant for subtype=4)
     73              0: symmetric square root
     74              1: Cholesky decomposition
     75
     76        Floating point parameters (Array param_real)
     77          param_real(1): Forgetting factor (usually >0 and <=1), required
     78
     79        Further parameters
     80          n_modeltasks: Number of parallel model integration tasks
     81              >=1 for subtypes 0 and 1; not larger than total number of processors
     82              =1 required for subtypes 2 and 3
     83          screen: Control verbosity of PDAF
     84              0: no outputs
     85              1: basic output (default)
     86              2: 1 plus timing output
     87              3: 2 plus debug output
     88
     89        Internal parameter (defined inside PDAF)
     90          Nm1vsN: Normalization of covariance matrix; default: 1
     91              0: normalization with 1/(Ensemble size)
     92                 (original SEIK, mainly for compatibility with older studies)
     93              1: normalization with 1/(Ensemble size - 1)
     94                 (sample covariance matrix consistent with other EnKFs)
     95
     96        +++++++++ End of option overview for the SEIK filter ++++++++++
     97}}}
     98
     99
     100== EnKF ==
     101
     102{{{
     103        Available options:
     104
     105        Sub-types (Parameter subtype)
     106          0: full ensemble integration; analysis for large dim_obs
     107          1: full ensemble integration; analysis for small dim_obs
     108          5: Offline mode
     109
     110        Integer parameters (Array param_int)
     111          param_int(1): Dimension of state vector (>0), required
     112          param_int(2): Ensemble size (>0), required
     113          param_int(3): maximum rank for inversion of HPH^T, required
     114              (if set to >=ensemble size, it is reset to ensemble size - 1)
     115
     116        Floating point parameters (Array param_real)
     117          param_real(1): Forgetting factor (usually >0 and <=1), required
     118
     119        Further parameters
     120          n_modeltasks: Number of parallel model integration tasks
     121              (>=1; not larger than total number of processors)
     122          screen: Control verbosity of PDAF
     123              0: no outputs
     124              1: basic output (default)
     125              2: 1 plus timing output
     126              3: 2 plus debug output
     127
     128        +++++++++ End of option overview for the EnKF ++++++++++
     129}}}
     130
     131
     132== LSEIK ==
     133
     134{{{
     135        Available options:
     136
     137        Sub-types (Parameter subtype)
     138          0: full ensemble integration; left-sided application of T
     139          2: Fixed error space basis
     140          3: Fixed state covariance matrix
     141          5: Offline mode
     142
     143        Integer parameters (Array param_int)
     144          param_int(1): Dimension of state vector (>0), required
     145          param_int(2): Ensemble size (>0), required
     146          param_int(3): not used
     147          param_int(4): 1 for incremental updating, 0 else; optional, default: 0
     148          param_int(5): Type of forgetting factor; optional, default: 0
     149              0: fixed forgetting factor
     150              1: adaptive forgetting factor for full domain (experimental)
     151              2: locally adaptive forgetting factor (experimental)
     152          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     153              0: deterministic omega
     154              1: random orthonormal omega orthogonal to (1,...,1)^T
     155              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     156                 (experimental; for random transformations, 1 is recommended)
     157          param_int(7): Type of transformation matrix square root; optional, default: 0
     158              (Only relevant for subtype=4)
     159              0: symmetric square root
     160              1: Cholesky decomposition
     161
     162        Floating point parameters (Array param_real)
     163          param_real(1): Forgetting factor (usually >0 and <=1), required
     164
     165        Further parameters
     166          n_modeltasks: Number of parallel model integration tasks
     167              >=1 for subtypes 0 and 1; not larger than total number of processors
     168              =1 required for subtypes 2 and 3
     169          screen: Control verbosity of PDAF
     170              0: no outputs
     171              1: basic output (default)
     172              2: 1 plus timing output
     173              3: 2 plus debug output
     174
     175        Internal parameter (defined inside PDAF)
     176          Nm1vsN: Normalization of covariance matrix; default: 1
     177              0: normalization with 1/(Ensemble size)
     178                 (original SEIK, mainly for compatibility with older studies)
     179              1: normalization with 1/(Ensemble size - 1)
     180                 (sample covariance matrix consistent with other EnKFs)
     181
     182        +++++++++ End of option overview for the LSEIK filter ++++++++++
     183}}}
     184
     185
     186== ETKF ==
     187
     188{{{
     189        Available options:
     190
     191        Sub-types (Parameter subtype)
     192          0: full ensemble integration; apply T-matrix analogously to SEIK
     193          1: full ensemble integration; formulation without T matrix
     194          5: Offline mode
     195
     196        Integer parameters (Array param_int)
     197          param_int(1): Dimension of state vector (>0), required
     198          param_int(2): Ensemble size (>0), required
     199          param_int(3): not used
     200          param_int(4): not used
     201          param_int(5): Type of forgetting factor; optional, default: 0
     202              0: fixed forgetting factor
     203              1: adaptive forgetting factor (experimental)
     204          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     205              0: deterministic transformation
     206              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     207
     208        Floating point parameters (Array param_real)
     209          param_real(1): Forgetting factor (usually >0 and <=1), required
     210
     211        Further parameters
     212          n_modeltasks: Number of parallel model integration tasks
     213              >=1 for subtypes 0 and 1; not larger than total number of processors
     214              =1 required for subtypes 2 and 3
     215          screen: Control verbosity of PDAF
     216              0: no outputs
     217              1: basic output (default)
     218              2: 1 plus timing output
     219              3: 2 plus debug output
     220
     221        +++++++++ End of option overview for the ETKF ++++++++++
     222}}}
     223
     224
     225== LETKF ==
     226
     227{{{
     228        Available options:
     229
     230        Sub-types (Parameter subtype)
     231          0: full ensemble integration;  apply T-matrix analogously to SEIK
     232          5: Offline mode
     233
     234        Integer parameters (Array param_int)
     235          param_int(1): Dimension of state vector (>0), required
     236          param_int(2): Ensemble size (>0), required
     237          param_int(3): not used
     238          param_int(4): not used
     239          param_int(5): Type of forgetting factor; optional, default: 0
     240              0: fixed forgetting factor
     241              1: adaptive forgetting factor for full domain (experimental)
     242              2: locally adaptive forgetting factor (experimental)
     243          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     244              0: deterministic transformation
     245              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     246
     247        Floating point parameters (Array param_real)
     248          param_real(1): Forgetting factor (usually >0 and <=1), required
     249
     250        Further parameters
     251          n_modeltasks: Number of parallel model integration tasks
     252              >=1 for subtypes 0 and 1; not larger than total number of processors
     253              =1 required for subtypes 2 and 3
     254          screen: Control verbosity of PDAF
     255              0: no outputs
     256              1: basic output (default)
     257              2: 1 plus timing output
     258              3: 2 plus debug output
     259
     260        +++++++++ End of option overview for the LETKF ++++++++++
     261}}}
     262
     263
     264== ESTKF ==
     265
     266{{{
     267
     268        Sub-types (Parameter subtype)
     269          0: Standard implementation with ensemble integration
     270          5: Offline mode
     271
     272        Integer parameters (Array param_int)
     273          param_int(1): Dimension of state vector (>0), required
     274          param_int(2): Ensemble size (>0), required
     275          param_int(3): not used
     276          param_int(4): not used
     277          param_int(5): Type of forgetting factor; optional, default: 0
     278              0: fixed forgetting factor
     279              1: adaptive forgetting factor (experimental)
     280          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     281              0: deterministic omega
     282              1: random orthonormal omega orthogonal to (1,...,1)^T
     283              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     284                 (experimental; for random transformations, 0 or 1 are recommended)
     285          param_int(7): Type of transformation matrix square root; optional, default: 0
     286              0: symmetric square root
     287              1: Cholesky decomposition
     288
     289        Floating point parameters (Array param_real)
     290          param_real(1): Forgetting factor (usually >0 and <=1), required
     291
     292        Further parameters
     293          n_modeltasks: Number of parallel model integration tasks
     294              >=1 for subtypes 0 and 1; not larger than total number of processors
     295              =1 required for subtypes 2 and 3
     296          screen: Control verbosity of PDAF
     297              0: no outputs
     298              1: basic output (default)
     299              2: 1 plus timing output
     300              3: 2 plus debug output
     301
     302        +++++++++ End of option overview for the ESTKF  ++++++++++
     303}}}
     304
     305
     306== LESTKF ==
     307
     308{{{
     309        Available options:
     310
     311        Sub-types (Parameter subtype)
     312          0: Standard implementation with ensemble integration
     313          5: Offline mode
     314
     315        Integer parameters (Array param_int)
     316          param_int(1): Dimension of state vector (>0), required
     317          param_int(2): Ensemble size (>0), required
     318          param_int(3): not used
     319          param_int(4): 1 for incremental updating, 0 else; optional, default: 0
     320          param_int(5): Type of forgetting factor; optional, default: 0
     321              0: fixed forgetting factor
     322              1: adaptive forgetting factor for full domain (experimental)
     323              2: locally adaptive forgetting factor (experimental)
     324          param_int(6): Type of ensemble transformation matrix; optional, default: 0
     325              0: deterministic omega
     326              1: random orthonormal omega orthogonal to (1,...,1)^T
     327              2: use product of 0 with random orthonomal matrix with eigenvector (1,...,1)^T
     328                 (experimental; for random transformations, 1 is recommended)
     329          param_int(7): Type of transformation matrix square root; optional, default: 0
     330              0: symmetric square root
     331              1: Cholesky decomposition
     332
     333        Floating point parameters (Array param_real)
     334          param_real(1): Forgetting factor (usually >0 and <=1), required
     335
     336        Further parameters
     337          n_modeltasks: Number of parallel model integration tasks
     338              >=1 for subtypes 0 and 1; not larger than total number of processors
     339              =1 required for subtypes 2 and 3
     340          screen: Control verbosity of PDAF
     341              0: no outputs
     342              1: basic output (default)
     343              2: 1 plus timing output
     344              3: 2 plus debug output
     345
     346        +++++++++ End of option overview for the LESTKF ++++++++++
     347}}}